INTERVIEW Jean Tirole

French economist Jean Tirole, recipient of the 2014 Nobel Memorial Prize in Economic Sciences, sees a close similarity between economics and the “caring profession” of medicine. “The economist,” he con- tended in his recent book Economics for the Common Good, “like the oncologist, makes a diagnosis on the basis of the best available (though necessarily imper- fect) knowledge, and then either proposes the most suitable treatment on that basis or no treatment at all, if none seems necessary.” The difference, for Tirole, is that in proposing policies, the economist must take into account the interests of people who aren’t in front of him or her and might not even be readily visible. “So the public sometimes accuses that economist of being indiffer- ent to the sufferings of the visible victims.” Tirole, who joined the Toulouse School of Economics in 1991 after a seven-year stint on the faculty of MIT, has made significant contributions to a wide array of areas within economics, including industrial organization, finance, banking regulation, and the economics of technology, to name a few. His 1988 textbook on industrial organization and his 2006 textbook on corporate finance are standards. With EF: How did you become interested in economics? Economics for the Common Good — his first book meant for popular audiences, published initially in French — Tirole: I was studying mathematics and physics in France. he seeks to bring the thinking and tools of academic In high school, I liked mathematics and social sciences. economists to a general readership. But my first class in economics wasn’t until I was 21 or 22 Ideologically, Tirole defies easy categorization. He at Ecole Polytechnique, which is an engineering school. It favors what he calls a “strong state” and a generous was a general introduction, actually quite a mathematical social safety net, but also argues for humility on the and theoretical one. Too much so indeed, because intu- part of regulators in light of the limited information ition and data are important in economics. Teaching eco- available to them. nomics is difficult — you want it to be rigorous but at the David A. Price interviewed Tirole in Washington, same time very intuitive, and there’s a certain trade-off. D.C., in November 2017. EF: And that one course was enough to move you in that direction?

Tirole: Yes, because I was attracted to the mix of the human aspect of the social sciences and the rigor of quan- titative analysis, a perfect combination for me. But I came late into the game, yes. I then got an applied math degree also, and finally I moved to MIT for a Ph.D., where I really learned economics.

EF: You said in your Nobel lecture that when you began studying industrial organization at MIT, you didn’t know what “industrial organization” meant.

Tirole: Yes. PHOTOGRAPHY: LISA HELFERT

22 E CON F OCUS | F OURTH Q UARTER | 2017 EF: What started you down So we have to take that into that road? We need to invent rules that are not account. Maybe not by breaking them up, because it’s hard to Tirole: It was totally fortuitous. too information intensive. Regulators break up such firms: Unlike for I was once in a corridor with don’t always have the required AT&T or power companies in my classmate Drew Fudenberg, information, so they need to have rules the past, the technology changes who’s now a professor at MIT. that are robust, that are going to work very fast; besides, many of the And one day he said, “Oh, there’s regardless of the circumstances. services are built on data that this interesting field, industrial are common to all services. But organization; you should attend to keep the market contestable, some lectures.” So I did. I took an industrial organization we must prevent the tech giants from swallowing up their class given by Paul Joskow and Dick Schmalensee, but not future competitors; easier said than done of course, as data for credit, and I thought the subject was very interesting are often missing to ascertain that the startup is indeed indeed. a competitor. And of course acquisitions are, along with I had to do my Ph.D. quickly. I was a civil servant IPOs, one of the standard routes for VCs and entrepre- in France. I was given two years to do my Ph.D. (I was neurs to cash out. granted three at the end.) It was kind of crazy. Bundling practices by the tech giants are also of con- cern. A startup that may become an efficient competitor EF: You’ve been credited along with others with cre- to such firms generally enters within a market niche; ating the first unified, coherent theory of industrial it’s very hard to enter all segments at the same time. organization. What are the main general policy pre- Therefore, bundling may prevent efficient entrants from scriptions that follow from your work in this area? entering market segments and collectively challenging the incumbent on the overall technology. Tirole: Both antitrust and regulation are hindered by Another issue is that most platforms offer you a best imperfect information. Regulators don’t have the infor- price guarantee, also called a “most favored nation” clause mation that managers of firms have. And for that reason, or a price parity clause. You as a consumer are guaranteed policymakers have to be humble. to get the lowest price on the platform, as required from This is a much broader lesson, by the way. For exam- the merchants. Sounds good, except that if all or most ple, in France, courts are involved in dismissal decisions, merchants are listed on the platform and the platform is so they have some say on whether the firm needs a given guaranteed the lowest price, there is no incentive for you to job or not. And of course, the court doesn’t have the look anywhere else; you have become a “unique” customer, information. You can make a similar observation with and so the platform can set large fees to the merchant to get “command-and-control” environmental policies. Both in access to you. Interestingly, due to price uniformity, these environmental and labor matters, governments sometimes fees are paid by both platform and nonplatform users — so suffer from hubris. They try to implement policies that each platform succeeds in taxing its rivals! That can some- they cannot implement efficiently because they don’t have times be quite problematic for competition. the information that’s required. Finally, there is the tricky issue of data ownership, I do believe in a strong state. But it requires regulation which will be a barrier to entry in AI-driven innovation. to be efficient; for that, governments must be humble and There is a current debate between platform ownership try to avoid intruding into an industry to do things that (the current state) and the prospect of a user-centric they simply cannot do. approach. This is an underappreciated subject that econo- mists should take up and try to make progress on. EF: One of the areas you address in Economics for the Common Good is the economics of online platforms EF: You’ve mentioned bundling, you’ve mentioned such as Amazon or Uber that bring together buyers most favored nation clauses — what should regulators and sellers. Do you think these platforms create spe- do about those things to keep markets contestable? cial issues for antitrust regulation? Tirole: It’s always difficult, and we need to make more Tirole: I think the answer is yes — partly because the new progress and do more research on simplifying the compe- platforms have natural monopoly features, in that they tition authorities’ task. Take bundling, for example. The exhibit large network externalities. I am on Facebook general lesson that you want to make the market contest- because you are on Facebook. I use the Google search able is fine, but sometimes bundling occurs for efficiency engine or Waze because there are many people using it, so reasons. And you have to look into the detail of each the algorithms are built on more data and predict better. case to see whether there’s a real efficiency difference or Network externalities tend to create monopolies or tight whether the firm is just trying to keep its competitors out. oligopolies. Another difficulty is that antitrust can be slow and

E CON F OCUS | F OURTH Q UARTER | 2017 23 digital industries are moving very Jean Tirole efficient search services, and so on. fast; if the authority decides too ➤ Selected Present Positions Of course there is a catch on the late, the entrant may already have Chairman of the Board, Fondation other side: the huge markups lev- folded. Another issue is territoriality Jean-Jacques Laffont/Toulouse Sciences ied on merchants or advertisers. But and possible disagreements among Economiques; Chairman, Executive we cannot just conclude from this authorities; a national competition Committee, Institute for Advanced observation that Google or Visa are authority may create a problem for Study in Toulouse;Visiting Professor, underserving monopolies on one side the incumbent because its decision Massachusetts Institute of Technology and are preying against their rivals on may force that firm to reconfigure its ➤ Education the other side. We need to consider products worldwide. Ph.D. (1981), Massachusetts Institute the market as a whole. We need to invent rules that of Technology; Diplôme d’Etude We have learned also that plat- are not too information intensive. Approfondie (1976), Docteur de 3ème forms behave very differently from Again, regulators don’t always have Cycle (1978), Université Paris-Dauphine; traditional firms. They tend to the required information, so they Engineer (1978), Ecole Nationale des be much more protective of con- need to have rules that are robust, Ponts et Chaussées; Engineer (1976), sumer interests, for example. Not that are going to work regardless of Ecole Polytechnique by philanthropy, but simply because the circumstances. We, for exam- ➤ Selected Publications they have a relationship with the ple, designed information-light rules “Deadly Embrace: Sovereign and consumers and can charge more to for patent pools; such rules enable Financial Balance Sheets Doom Loops,” them (or attract more of them and antitrust authorities to say, “OK, we Review of Economic Studies, forthcoming cash in on advertising) if they enjoy a allow you to form a patent pool if (with Emmanuel Farhi); Economics for higher consumer surplus. That’s why you meet such, such, and such con- the Common Good (Princeton University they allow competition among appli- ditions.” These rules — (a) patent Press, 2017); “Efficient Patent Pools,” cations on a platform, that’s why owners should keep ownership of American Economic Review, 2004 (with they introduce rating systems, that’s the patents and thus be able to grant Josh Lerner); “Intrinsic and Extrinsic why they select out nuisance users Motivation,” , individual, outside-the-pool licenses, Review of Economic Studies (a merchant who wants to be on the 2003 (with Roland Bénabou); and (b) the pool should make unbun- The Theory platform usually has to satisfy various of Industrial Organization (MIT Press, dled offers for the licenses — require 1988); and numerous other articles and requirements that are protective of no information from the antitrust books consumers). Those mechanisms — authorities. for example, asking collateral from I think we need to do more rule participants to an exchange or put- design to facilitate the antitrust authorities’ work, because, ting the money in an escrow until the consumer is satisfied even leaving aside the financial cost of collecting, verifying, — screen the merchants. The good merchants find the and analyzing data, authorities cannot afford to spend five cost minimal, and the bad ones are screened out. or 10 years deciding, right? Besides, products that are com- That’s very different from what I call the “vertical plements today may become substitutes tomorrow, or the model” in which, say, a patent owner just sells a license opposite. Because the usage changes, the competitive pat- downstream to a firm and then lets the firm exercise its tern changes. The job of antitrust authorities is extremely full monopoly power. difficult in the end and we economists have to help them. I’m not saying the platform model is always a better model, but it has been growing for good reason as it’s EF: Your research with Jean-Charles Rochet started more protective of consumer interest. Incidentally, today a whole new literature of two-sided markets. This the seven largest market caps in the world are two-sided has been influential on both industry participants platforms. and policymakers with regard to platform industries, But there is of course the other side, which is the especially those related to payments. What do you merchant interest. So the right balance has to be found, think is the main lesson for people to take away from and both platforms and antitrust authorities are trying that research? to do so.

Tirole: Both authorities and private decisionmakers must EF: In that respect, it sounds like these platforms can analyze the two sides at the same time. For example, for be regulators themselves. Is that a concern? competition policy in the payment card industry, authori- ties cannot just look at the merchant side or the cardholder Tirole: I’m not too worried about that. I see antitrust side. They have to look at the interaction between both. issues with the platforms — they can do the wrong thing We get a fantastic deal from Google or credit card socially — but these regulatory activities don’t look so platforms. Their services are free to consumers. We get bad because they try to avoid dissipation of total surplus. cashback bonuses, we get free email, Waze, YouTube, Without giving them a blank check, I think that this

24 E CON F OCUS | F OURTH Q UARTER | 2017 particular activity of platforms is quite useful for society. objectionable (is an American worker really more deserv- ing than an Indian one?), and that it may hurt consumers EF: Predictions of massive unemployment or under- both directly (reduced access to what the world has to employment from automation have been common for offer) and indirectly (by creating domestic monopolies a long time. As the tech industry is creating more dis- that will raise prices and slow down innovation). intermediation and bringing new kinds of automation In Europe, especially southern Europe, protecting jobs to the workplace, are you bullish about the future of would seem the favorite option. But there’s only so much jobs that pay good wages? you can do to protect jobs: First, you’re only slowing down the adjustment, and second, it means that no stable job Tirole: History tells us that there is never a shortage of jobs. is created anymore, leaving the scene to gig jobs. If you People have been predicting for two centuries that there protect the job too much through labor laws, as in France, will be, but in the end there are always new jobs to meet what happens of course is that employers respond. So new needs for consumers and provide new services that can 90 percent of the jobs that are created today in France be supplied. So I’m not concerned, per se, about a shortage are temporary jobs. Tomorrow it will be 95 percent if no of jobs. I’m more concerned about a shortage of jobs that reform is made. It’s inefficient. Temporary jobs are bad people will want to take. The danger right now is that the jobs. You’re employed for a month, three months, and then jobs that are likely to be created may be low-paying ones. you go through the unemployment spell, and then you are The losers, not only of globalization, but also of techno- rehired. And employers don’t invest in your human capital logical progress, either saw their wages stagnate over the — you are perceived as the equivalent of a disposable tissue. last 30 years or ended up being unemployed (in southern So we can slow down the adjustment, but in the end, Europe), or underemployed, or employed in gig jobs (in we do have to protect workers in a different way. The the U.S. and the U.K). Some of my colleagues have doc- Scandinavians protect workers through generous unem- umented a polarization between lower-paid people, who ployment benefits; at the same time, the latter are required have low skills and haven’t seen their salaries increase (or to work hard to find a new job. Workers are both well have in some cases seen their salaries decrease), and of protected, as they should be — it’s usually not their fault if course the high-skilled people who have benefited greatly they lose their job — and they are made accountable, in that from globalization and technological progress. they must search hard for a job and take an appropriate job This may well keep happening. But not only to the if available. It’s a quid pro quo. This Scandinavian contract low-skilled workers: AI also threatens the jobs of highly is no panacea, but it is probably the best we have. skilled people. For example, the role of doctors is going to In the U.S., there has not been enough protection of be different. Here I’m not talking about the MIT biotech- the losers of globalization and technological change. That nologist or Harvard medical school professor, I’m talking discontent was reflected in the last election. It’s not the about general practitioners and the like. only explanation, of course, because we also see in Europe It’s going to be hard for many. Take teachers. Or law- countries that are doing well despite votes for populist yers: Algorithms already do part of the lawyer job. Not all movements being high. Populism has broader causes, but of it; writing a convincing legal argument, for example, discontent and anxiety about the future do not help. is still very difficult, but the identification of all the rel- The education system in many countries is not up evant cases and the preparatory work can be done by an to scratch. Just to take the example of France, the top algorithm and it can be done very well. So it’s not only the 20 percent of students are very well educated, and those low-skill jobs I’m worried about. will again be the winners in the new world. But the 80 per- Less-developed and emerging countries that are trying cent below actually don’t get a good education. Later on, to develop markets for their cheap labor — which has they will get poor vocational retraining, even though we done wonders in the past for China and India — will have spend 31 billion euros per year on this. You can understand to adapt their strategy as their flagship jobs, such as those people being very worried about their future. The growth in call centers, are going to become obsolete. in inequality is not likely to subside with such policies. Jobs will be destroyed faster and faster, requiring more While I’m confident that we’ll be overall much richer worker protection and less job protection. Of course, edu- and healthier, I’m pessimistic about our social compact if cation and retraining will be key. We are not yet ready for we don’t react to the challenge. providing it efficiently. I don’t have any miracle cure, but it’s going to be a big issue. EF: If we could shift gears again, you’ve made import- What are the alternatives? Some propose some form of ant contributions to banking theory. What do you protectionism against imports of goods and services. That think are the most important lessons of the financial would appear to be the current trend in the U.S.; we’ll crisis for the further development of banking theory? see. We should however not forget that protectionism can be self-defeating (retaliation by other countries hurts Tirole: Some of the lessons of the crisis we knew before- workers in exporting industries) and possibly morally hand. We knew, for example, that large over-the-counter

E CON F OCUS | F OURTH Q UARTER | 2017 25 positions or capital requirement evasion through off- regulate anything you deem “too big to fail” or “too sys- balance-sheet special-purpose vehicles, like the conduits temic to fail.” That’s the systemically important financial associated with securitization, can be factors of financial institution, or SIFI, approach. It’s fine, but the question instability. But we had no clue about the actual magni- is, as I said, how do you identify “too big to fail”? The size tude of these arrangements — at least people in academia of the balance sheet and the leverage ratio are informative, didn’t. Maybe economists who were closer to banking but imagine that you face a large financial institution that supervisors and financial institutions may have known. invests only in safe assets. Is it “too big to fail”? No. So size Better regulatory infrastructure is very important. And per se is not very informative. One must dig deeper into I think we have made progress. In the U.S., regulatory the risk that it could create for financial stability, not an forum shopping, where banks could choose their regula- easy task for a regulated entity and even less so for a previ- tor, has been made more difficult. Some regulators were ously unregulated one. Conversely, you can have a smaller very lenient. So having one player only — the Fed, which institution whose failure would create a lot of trouble. has some independence and credibility to do the supervi- As Emmanuel Farhi and I described in a recent NBER sion -- was a good thing, in my view. Similarly, increasing paper, a shadow bank may be bailed out for two reasons. the distance between banks and their supervisors through One is the threat of financial contagion that its failure the single supervisory mechanism in Europe was at least in might engender, as we just discussed. The other is that it theory a good move. serves what I call “politically fragile” clients. On the liabil- Did we learn all these things with the crisis? No. But ity side, it will be small depositors; on the asset side, it will did we start putting more emphasis on them because of be small and medium enterprises. And if those politically the crisis? Yes, I think that’s correct. Partly because, as I fragile clients migrate to the shadow banking sector, as say, it’s not only the regulators who have imperfect infor- happens in China, for example, and more and more in mation; the economists, at least those in academia, also Europe and the U.S., the state might actually bail out the are imperfectly informed. They can say that a practice is shadow banks because it wants to protect them. dangerous, but as long as they don’t know whether it is widespread or limited, they won’t spend that much time EF: When you say “politically fragile,” what do you warning about it as they can’t document a concern; neither mean? will supervisors take the advice seriously. Let me provide another illustration. Shadow banking Tirole: If small depositors lose their savings, there’s a is attracting substantial academic attention nowadays. strong temptation to intervene and make sure they don’t, Economists are trying to make progress in understanding because that’s all they have. The same goes for small and the role of shadow banking and what kind of danger it medium enterprises, which may not be as resilient as larger presents. Again, what should we do, taking into account firms: The state might be concerned that if the SMEs lose the fact that regulators have limited information? Even their lender, which has specific information about them if regulators are independent of industry and political and engages in relationship banking with them, then eco- power, they still need the information to do a good job. nomic activity may suffer. That’s one reason why shadow We always come back to the same thing: We have to banks may be rescued. design rules that are not too intrusive and basically work The other reason is more related to 2008, in that with the actual information that regulators have. there might be problematic cross-exposures as I discussed previously. Regulators were worried that with AIG going EF: One problem that was obviously important in the under, regulated banks or insurance companies that would financial crisis is the so-called “too big to fail” problem. have lost money through their cross-exposures with AIG What do you think is the best solution to this problem, might themselves get into trouble. where regulators know that the possibility of a bailout So there are two ways of addressing the shadow banking will work against the firms’ self-discipline, but regula- problem. As I mentioned, one is to declare X as a system- tors may still feel they need to provide a bailout when a ically important institution and to supervise this financial crisis occurs to avoid letting it become worse? institution and require capital adequacy from it. The alternative approach is some kind of ring fencing, which Tirole: Well, that’s correct. One issue is that “too big to is meant to prevent regulated banks from being exposed fail” is always difficult to define. Was LTCM [Long-Term to the possible failure of a shadow bank: One tries to keep Capital Management] truly too big to fail? Was AIG? the unregulated sphere away from the regulated one. I It’s difficult because the potential for financial contagion think the debate now is between those two competing depends on the troubled institution’s balance sheet, on philosophies, which both have their advocates. the correlation of exposures, on who is on the other side, and so on. EF: Many of us in America think of Paris having a role There are two possible strategies and a lively debate in France similar to that of New York and Washington among economists about those two strategies. One is to combined in the United States. Do you feel it’s been

26 E CON F OCUS | F OURTH Q UARTER | 2017 advantageous for you to be based away from Paris — economist. I was there as a student, I was there on the more or less on the opposite side of the country? faculty, and I’ve stayed there as a visiting professor for 26 years now. Today there are many new faces, but the culture Tirole: That’s right, Paris is both the political and eco- has been preserved. nomic capital. France is really centered around Paris. And I work hard, but everyone works hard. My one merit is traditionally many of the good schools and universities are to have been with the right people. You meet people who in Paris, especially in the human and social sciences. change your life, and they are your eye-openers. I had no personal reason to be in Toulouse except for my very charismatic friend, Jean-Jacques Laffont, who started EF: Your book is written very clearly, very differently a top economics department there. He was one of the top from academic economic research. What were you economists in the world, at the same time he spent much trying to achieve in writing the book? time developing something in what was at the time a rather unlikely place, and I admired that. That’s why I went to Tirole: In the past, I had been engaging with experts — in Toulouse in 1991, and that’s why I stayed in Toulouse. corporations, in government, in regulatory agencies, in But if you look at the U.S., the top universities may be central banks, and so on. But I never actually interacted in remote places. Or they may be in places like Boston with a wider audience. I defined my mission as research or Chicago that are big cities but are not the economic and teaching. But then with the Nobel Prize, people or political capital. In my view, there’s no reason to be started asking me questions, mostly about the work of in the capital unless you want to constantly advise the economists, what do they do, are they useful, what’s their government. methodology, and so on. So I thought it would be useful to And even then, I’ve been on the Council of Economic write a book to try to explain. Advisers of France for almost 20 years. The Council of Now I go to high schools, which I didn’t used to do. I Economic Advisers is a nonpartisan body, so I’ve been serv- talk to people, I give speeches for wider audiences, which ing a number of right-wing and left-wing governments. I’ve again I didn’t do before. The prize was the wakeup call. also belonged to many committees in Paris. From Toulouse, And in retrospect, the timing was right in light of the I can still perform public service at the national level. populist movements we have discussed earlier that are But for research and teaching, there’s absolutely no all over the world. At some point, if the population as a reason for why you have to be in the center of political whole doesn’t take economics, and science more generally, decision-making. It’s not the tradition in France, but the onboard — these thoughts would alternatively apply to U.S. and other countries have shown that you can have medicine, evolution, biology, or climate science -- if we really top universities away from the centers of decision. don’t manage to pass on basic knowledge, it’s very hard for democracy to work. We get the policies that we deserve. EF: Who would you say have been your main So we have to educate people in basic knowledge. Of influences? course we cannot ask people to have a Ph.D. in economics or medicine or biotechnology, but we can provide them Tirole: You are what your collaborators, and more generally, with the basic knowledge to think about what is a fact, an colleagues, make of you. They bring the best out of you, and empirical test, the difference between a correlation and a they bring their own contribution; it’s a collective endeavor. causality, the nature of a theory, how to avoid pitfalls in There are two people who have influenced me more directly. reasoning. One is Jean-Jacques Laffont, both through our joint work All of those things could be taught in high school to and through his being a role model — someone actually some extent. We academics need to share better knowl- working a lot for the common good. He never put his career edge within the population, because in the end, politicians, first, which did not prevent him to have a very distinguished and I’m not blaming them, tend to focus on reacting to career until cancer took him away in 2004; he was obsessed what the electorate wants. with building a top institution in Toulouse, helping less-de- If you can advise governments, that’s useful, but at some veloped countries, and many other things. The other is Eric point, if you don’t also have an adoption by the electorate Maskin, 2007 Nobel Prize laureate, who is similar in many as well, the policies that you’ll get won’t be the right ones ways and was incidentally a close friend of Jean-Jacques. He necessarily. We have to rehabilitate, we have to create was my adviser at MIT, and I owe him a lot. more trust and faith in experts. Sometimes those experts MIThas been a special institution for me, including the can be blamed, too. Our judgment may be impaired. But if “old guard”: Bob Solow, Olivier Blanchard, Stan Fischer, the population has no respect for experts, anything goes, Paul Joskow, Paul Samuelson, Peter Diamond — I’m not right? Anything goes, and then you may end up with bad going to name all my mentors and friends, but MIT has outcomes. been a very decisive influence in my career. Learning to On my small scale, I’m trying to do something. On work by this combination of both a rigorous approach my small scale. But if we all do that, we can improve and intuitive thinking has been important to me as an things. EF

E CON F OCUS | F OURTH Q UARTER | 2017 27